Comparing the Influence of Phonological Network Structure on Spoken Word Recognition Performance Across Speakers of North American and Singapore English

The mental lexicon is a repository of all known words and their phonological connections in long-term memory. These connectivity patterns can be visualized through phonological networks, with network metrics (degree and local clustering coefficient) having previously been observed to influence spoken word recognition. However, it remains unclear whether different dialects of a language have distinct phonological networks and whether such differences affect cross-dialect word recognition. This study compared American English (AmE-Net) and Singaporean English (SgE-Net) phonological networks on predicting word detection performance of native speakers of American English and Singaporean English, for words spoken in both dialects. We hypothesized that network metrics from a participant’s dialect would better predict their spoken word recognition in their own dialect. Results were not entirely as expected: The pattern of the interaction effects suggested that AmE-Net degree was the superior predictor both participant groups; yet, SgE-Net degree, but not AmE-Net degree, was a significant predictor when words were produced by the Singaporean talker. The Singaporean mental lexicon may thus be more influenced by AmE than previously anticipated. Overall, phonological networks remain valuable for modeling dialect differences, though their predictive power may depend on listener familiarity with the dialect.